I am trying to calculate the mean distance between nearest neighbor of a series of GPS points in R. I found two codes to get the values. But it doesn't appear to give the correct distance in meters. When I check against Google Maps, its' way off.
I found this answer: R - Finding closest neighboring point and number of neighbors within a given radius, coordinates lat-long
library(geosphere)
sp.groups <- groups
coordinates(sp.groups) <- ~Long+Lat
class(sp.groups)
d<- distm(sp.groups)
min.d<- apply(d, 1, function(x) order(x, decreasing=F)[2])
min.d
mean(min.d)
groupdist<- cbind(groups, groups[min.d,], apply(d, 1, function(x) order(x, decreasing=F)[2]))
colnames(groupdist)<- c(colnames(groups), 'neighbor', 'n.lat', 'n.long','dist')
And here using the package rgeos, but it gives the same results: Calculate the distance between two points of two datasets (nearest neighbor)
library(rgeos)
sp.groups <- groups
coordinates(sp.groups) <- ~Long+Lat
proj4string(sp.groups) <-CRS("+proj=utm +datum=WGS84")
class(sp.groups)
d<- gDistance(sp.groups, byid=TRUE)
min.d<- apply(d, 1, function(x) order(x, decreasing=F)[2])
min.d
mean(min.d)
groupdist<- cbind(groups, groups[min.d,], apply(d, 1, function(x) order(x, decreasing=F)[2]))
colnames(groupdist)<- c(colnames(groups), 'neighbor', 'n.lat', 'n.long','dist')
When I go and check on Google Earth the distances can be way off. And it even gives different values for nearest neighbors by 160-200m. Also some of the nearest neighbors don't have the same distance values, see K11 and K3 then K3 and K11. Here's the results I get, and I added expected values from Google Maps:
Group Lat Long neighbor n.lat n.long dist GMaps
K1 -26.96538 21.80965 K34 -26.96503 21.80940 27 44
K10 -26.96575 21.81132 K1 -26.96538 21.80965 1 172
K11 -26.96249 21.81120 K3 -26.96387 21.81053 22 166
K24 -26.96033 21.81090 K11 -26.96249 21.81120 3 240
K3 -26.96387 21.81053 K11 -26.96249 21.81120 3 166
K34 -26.96503 21.80940 K1 -26.96538 21.80965 1 44
What is wrong?
My data
groups<-data.frame(Group = c('K1', 'K10', 'K11', 'K24', 'K3', 'K34'),
Lat = c(-26.96538, -26.96575, -26.96249, -26.96033, -26.96387, -25.96503),
Longitude = c(21.80965, 21.81132, 21.81120, 21.80190, 21.81053, 21.80940))
I am not sure what you tried to calculate here but the column distance just refers to the position/number of the point with the minimum distance. I added a number with the actual minimum distances which look like the result you expected.
library(geodist, include.only = NULL)
library(sp, include.only = NULL)
groups <- data.frame(Group = c('K1', 'K10', 'K11', 'K24', 'K3', 'K34'),
Lat = c(-26.96538, -26.96575, -26.96249, -26.96033, -26.96387, -25.96503),
Long = c(21.80965, 21.81132, 21.81120, 21.80190, 21.81053, 21.80940))
sp.groups <- groups
sp::coordinates(sp.groups) <- ~Long+Lat
# mindistance Matrix
d <- geodist::geodist(groups, measure = "cheap")
# position of minimum distance
diag(d) <- Inf
min.d <- max.col(-d)
min.d
#> [1] 2 1 5 5 3 4
groupdist <- cbind(groups, groups[min.d,], min.d)
colnames(groupdist) <- c(colnames(groups), 'neighbor', 'n.lat', 'n.long','closest_to')
# get minimum distance for each pair of coordinates
groupdist$distance <- d[cbind(seq_along(min.d), min.d)]
groupdist
#> Group Lat Long neighbor n.lat n.long closest_to distance
#> 2 K1 -26.96538 21.80965 K10 -26.96575 21.81132 2 171.1554
#> 1 K10 -26.96575 21.81132 K1 -26.96538 21.80965 1 171.1554
#> 5 K11 -26.96249 21.81120 K3 -26.96387 21.81053 5 167.2225
#> 5.1 K24 -26.96033 21.80190 K3 -26.96387 21.81053 5 944.4119
#> 3 K3 -26.96387 21.81053 K11 -26.96249 21.81120 3 167.2225
#> 4 K34 -25.96503 21.80940 K24 -26.96033 21.80190 4 110613.1362
Created on 2021-08-22 by the reprex package (v2.0.1)